基于图像处理的尿液流量测量系统

An urine-flow measuring system based on image processing

为了实现对危重病人尿流流量的检测,提出了一种基于图像处理的尿液流量测量系统。与工业上的液位检测系统相比,该系统采集到的图片中的液位不是很明显,但又需要液位检测过程简单高效。鉴于以上的矛盾,采用二值化检测储液盒底部和顶部刻度线,然后将底部线和顶部线之间的区域提取出来做接下来的运算。不仅减少运算量还对储液盒进行定位,提高了精度。根据图片的特点,对运算区域依次进行直方图均衡、sobel水平边缘检测、液位提取,最终得到液位值。对于计算过程中需要的阈值,通过pc端的智能参数筛选程序利用遗传算法得到。试验证明使用遗传算法迭代500次后得到的阈值能够使系统的误差小于3个像素,表明通过图像处理的方法能够进行尿液流量的测量。

This text raises an urine-flow measuring system based on image processing to measure urine-flow of sick persons who be in a grave condition.Compared with urine-flow measuring system in industry,level photos collected by this system lack definition,and liquid_level needs to be detected simply and effficiently.To the contradiction,this text used binarization method to detect the graduation marks at the bottom and top of the reservoir,then extracted area between the bottom line and the top line for later computation,which is not only reduced computation,but also localized the reservoir and improved precision. According to characteristics of photos, successively used histogram equalization, sobel horizontal edge detection,extracting liquid level to get liquid-level value. Used a PC program which uses genetic algorithm （GA） to intelligently select parameters to get values that are needed in the later computations. The result shows that values which are got from GA that has 500 iterations let error value less than 3 pixels,and use image processing can measure flow of urine.